Megapixels?

[al-alami]

Hi Everyone,

As a working photographer I'm used to having high megapixel cameras, when looking at good astro ccds I find that megapixels are well below what I expect.  Please be patient with me I'm new at astrophotography.  I have a canon 1200D which I moded with the baader internal filter.  If my aim is to ultimately make large prints and I want good quality, I know (and understand) the need for a cooled mono CCD.  But why are the numbers so low?

If I intend on going narrowband, does that mean if I use a 2 megapixel ccd its 2 megapixels per layer? As i 2 for Ha, 2 for OIII and 2 for SII? So the picture would be 6 megapixels in size?

I guess you can tell that I am a bit confused by this, and I can't seem to find the right answer anyway (I did try looking, honest).

Thanks in advance,

Fakhri

[furrysocks2]

I have little to no experience but have been researching this recently. I'm sure others will be along shortly to give you more knowledgeable answers.

 

Make sure to consider pixel size, sensor size and image scale when making comparisons. Note that there are different rules of thumb regarding sampling rate and arc-seconds per pixel for DSO and planetary imaging, for example. Also worth understanding binning.

Your narrowband (mono) exposures would typically be combined by assigning them to R, G or B channels (for example, the "Hubble Palette"), yielding a colour image at the same resolution. Worth picking up a copy of the book "Making Every Photon Count".

[ollypenrice]

It's a good question and the answer is straightfoward in principle. There comes a point at which the pixels of the camera are so small as to be trying to resolve fine detail below the level at which the telescope, mount and atmospheric stability can deliver it onto the chip. At this point you are wasting your time. Atmospheric turbulence and guiding error will blur the detail before it gets to the camera so you might as well use bigger pixels which receive more light. You will lose no detail that way. (If you were using a space based telescope with perfect tracking you could usefully use much smaller pixels because the atmosphere wouldn't affect the image.)

Everyone will be happy with that very standard explanation as far as it goes. (I think!  ) It becomes a hotly debated topic when you try to decide at what point it is pointless to image at higher resolution and here there is no consensus. Firstly a lot depends on your local seeing conditions - and these very enormously from night to night and even within one night. The unit of resolution in question is arcseconds per pixel. Plenty of online calculators will give you this. Out of habit I use this one: http://www.12dstring.me.uk/fovcalc.php

It is easier to asses the upper limit of "PP than the lower. Small stars will start to have a blocky or square look if you go much above 3.5"P/P. By 4.46"P/P I found I could still get acceptable results but dense fields of small stars sometimes needed the stars gently blurring. At 3.5"PP we get perfectly shaped stars but not the detail the telescope could resolve with smaller pixels.

At the high res end you won't find much agreement! Some think the Nyquvist theorem applies, some don't. The highest resolution claims that are generally made are for sampling rates of about 0.5"P/P but these are controversial. Many imagers call it a day at 1 arcsecond per pixel. Personally that sounds about right to me. You have to bear in mind that so many obstacles stand between the imager and a good image that taking on too much just adds to them.

Don't forget that CCD chip sizes vary enormously so you need to look at pixel size and chip size. Nobody makes large CCD chips with small pixels, unfortunately. At least not for astronomy.

Olly

[johnrt]

1 hour ago, al-alami said:

If I intend on going narrowband, does that mean if I use a 2 megapixel ccd its 2 megapixels per layer? As i 2 for Ha, 2 for OIII and 2 for SII? So the picture would be 6 megapixels in size?
 

A comprehensive answer from Olly, but this part of your post caught my attention. The answer is no, the images from each filter are assigned to RGB channels in Photoshop to make a colour image, this does not increase the size or resolution of your image, a 2 mega pixel chip produces a 2 mega pixel image no matter how many filters have been used to construct it.

The difference between the DSLR chip and the mono CCD is that the DSLR chip has a bayer marix over it so never uses all it pixels to capture each colour, using a mono CCD and filters lets you capture each colour in turn and use all the pixels to do so, giving you a much better signal to noise ratio, but it is still always remains a 2 mega pixel image

[al-alami]

53 minutes ago, ollypenrice said:

It's a good question and the answer is straightfoward in principle. There comes a point at which the pixels of the camera are so small as to be trying to resolve fine detail below the level at which the telescope, mount and atmospheric stability can deliver it onto the chip. At this point you are wasting your time. Atmospheric turbulence and guiding error will blur the detail before it gets to the camera so you might as well use bigger pixels which receive more light. You will lose no detail that way. (If you were using a space based telescope with perfect tracking you could usefully use much smaller pixels because the atmosphere wouldn't affect the image.)

Everyone will be happy with that very standard explanation as far as it goes. (I think!  ) It becomes a hotly debated topic when you try to decide at what point it is pointless to image at higher resolution and here there is no consensus. Firstly a lot depends on your local seeing conditions - and these very enormously from night to night and even within one night. The unit of resolution in question is arcseconds per pixel. Plenty of online calculators will give you this. Out of habit I use this one: http://www.12dstring.me.uk/fovcalc.php

It is easier to asses the upper limit of "PP than the lower. Small stars will start to have a blocky or square look if you go much above 3.5"P/P. By 4.46"P/P I found I could still get acceptable results but dense fields of small stars sometimes needed the stars gently blurring. At 3.5"PP we get perfectly shaped stars but not the detail the telescope could resolve with smaller pixels.

At the high res end you won't find much agreement! Some think the Nyquvist theorem applies, some don't. The highest resolution claims that are generally made are for sampling rates of about 0.5"P/P but these are controversial. Many imagers call it a day at 1 arcsecond per pixel. Personally that sounds about right to me. You have to bear in mind that so many obstacles stand between the imager and a good image that taking on too much just adds to them.

Don't forget that CCD chip sizes vary enormously so you need to look at pixel size and chip size. Nobody makes large CCD chips with small pixels, unfortunately. At least not for astronomy.

Olly

Oh wow ... Thanks Olly .. I think. 

I think I will have to read that quite a few times before I understand it totally. 

To think, back in the day we used to confider shooting with slides on a 4x5" view camera to be the most technically challenging form of photography.  I stand very well corrected. 

I think I am going to enjoy being an amateur photographer again. 

Lots and lots and lots to learn.

I guess the best way to do all this, is make the best out of my DSLR so that I learn the ropes properly, and work myself up to the right scope and ccd combo.  It seems that you can't buy a ccd that works with "any" scope.  (not to mention the fact that ccd cameras are not cheap, and that considering I live in Jordan, I would have to have anything shipped to me from Europe, so add shipping and customs)

But I do appreciate all the info

Fakhri

 

[al-alami]

48 minutes ago, johnrt said:

A comprehensive answer from Olly, but this part of your post caught my attention. The answer is no, the images from each filter are assigned to RGB channels in Photoshop to make a colour image, this does not increase the size or resolution of your image, a 2 mega pixel chip produces a 2 mega pixel image no matter how many filters have been used to construct it.

The difference between the DSLR chip and the mono CCD is that the DSLR chip has a bayer marix over it so never uses all it pixels to capture each colour, using a mono CCD and filters lets you capture each colour in turn and use all the pixels to do so, giving you a much better signal to noise ratio, but it is still always remains a 2 mega pixel image

Why isn't it that the simple logical answer isn't correct! 

I've actually read a lot about the bayer matrix, and people who have removed it from their dslrs, not sure if I have the guys to even try that.  

[ollypenrice]

If you want to start with your DSLR I suggest you pick a nice short focal length telescope to go with it, something giving you maybe 2 to 4 arcsecs per pixel. You probably have good camera lenses so why not start with those? Then you could master the business of polar aligning and autoguiding.

A mono CCD can be slightly flexible in that, if the pixels are too small, you can bin them 2x2. This makes a virtual pixel four times larger by area. You can't do this with a Bayered chip because the colour information would be destroyed. (The matrix would be over-ruled by the binning.)

Olly

[al-alami]

5 hours ago, ollypenrice said:

If you want to start with your DSLR I suggest you pick a nice short focal length telescope to go with it, something giving you maybe 2 to 4 arcsecs per pixel. You probably have good camera lenses so why not start with those? Then you could master the business of polar aligning and autoguiding.

A mono CCD can be slightly flexible in that, if the pixels are too small, you can bin them 2x2. This makes a virtual pixel four times larger by area. You can't do this with a Bayered chip because the colour information would be destroyed. (The matrix would be over-ruled by the binning.)

Olly

I've got a Skywatcher 130PDS, I guess you could call that a nice short focal length .... right?

If I'm reading this right (from the link you included about its 2.04arcsecs.

[Owmuchonomy]

Olly makes a good point considering your experience. I did exactly that and took my first deep sky picture with a DSLR and a 300mm lens.  I learnt an awful lot before expanding my range of kit.  On the topic of a 130P DS I have just sold mine but it is in the right hands a very good tool considering the price point. What that combination will do is focus your attention on achieving a full flat field on a DSLR chip and dealing with coma and lack of in focus. The journey begins!

[al-alami]

2 minutes ago, Owmuchonomy said:

Olly makes a good point considering your experience. I did exactly that and took my first deep sky picture with a DSLR and a 300mm lens.  I learnt an awful lot before expanding my range of kit.  On the topic of a 130P DS I have just sold mine but it is in the right hands a very good tool considering the price point. What that combination will do is focus your attention on achieving a full flat field on a DSLR chip and dealing with coma and lack of in focus. The journey begins!

Its a very steep learning curve, but its fun (well, when you have the right knowledge and equipment  ) or at least so far

Given that I live in a turbulent part of the world, people tend to forget to look up, and even with the few pictures I have been able to take so far, everyone is like wow.  Yet maybe being on a forum like this, I've seen results a million times better.  Of course I never expected to do great from the beginning, but so far I'm enjoying the experience of learning again

Maybe I have high expectations of myself